Forecasting the pollution load of non-point sources to the Jiuzhou River

摘要

Based on the investigation of Jiuzhou River, and puts forward the new method based on GP about forecast non-point source pollutants. Jiuzhou river since 1986-1995 which the monitoring data as training samples and testing samples. Nonpoint source pollution load and its influencing factors of the nonlinear mapping relationship between, can be achieved by GP structure learning and training samples. The monitoring data of Jiuzhou River since 1996 to 1998 are preformed to testify the effects of the method above. Compared with other machine learning methods, it show that GP method is more feasible, effective and simple.